Server specifications for a single OBServer node
The following table describes the server specifications required for a single OBServer node to ensure stable cluster operation.
Configuration |
Maximum Requirement |
Minimum Requirements |
|---|---|---|
| CPU | 512C | 4C |
| Memory | 1 TB | 16 GB |
Cluster name length limit
Data Item |
Maximum Length |
|---|---|
| Cluster Name | 128 bytes |
Identifier length limit
Item |
Maximum length |
|---|---|
| Username | 64 bytes |
| Tenant name | 63 bytes |
| Database name | 128 characters |
| Table name | 64 characters |
| Partition name | 64 characters |
| Column name | 128 characters |
| Index name | 64 characters |
| View name | 64 characters |
| Alias | 255 characters |
| Table group name | 127 bytes |
| User-defined variable | 64 characters |
Item |
Maximum length |
|---|---|
| Username | 64 bytes |
| Table name | 128 bytes |
| Column name | 128 bytes |
| Partition name | 64 bytes |
| Index name | 128 bytes |
| View name | 128 bytes |
| Alias | 128 bytes |
| Object name | 128 bytes |
| Table group name | 127 bytes |
Maximum number of tenants that can be created on a single OBServer node
Type |
Maximum Limit |
|---|---|
| Number of tenants that can be created on a single OBServer node | The number of tenants is affected by the cluster deployment architecture and hardware factors. OceanBase Database supports a minimum tenant memory of 4 GB, which can be calculated as one tenant per 4 GB. |
Connection limits for OceanBase Database Proxy (ODP)
Maximum number of connections
Type |
Maximum Limit |
|---|---|
| Number of connections per ODP | Controlled by ODP's client_max_connections parameter. The default value is 8192.
NoteYou can increase the total number of connections in a cluster by adding ODP nodes or modifying the value of the |
Connection limits per OBServer node
Type |
Maximum Limit |
|---|---|
| Number of Connections per OBServer Node | 256K |
Number of partition replicas
Type |
Maximum Limit |
|---|---|
| Number of Partition Replicas per OBServer Node | No strict limit.
NoteYou can estimate the number of partition replicas for each OBServer node based on the tenant's memory size. Approximately 1 GB of memory supports about 20,000 tablets. |
Limits for a single table
Type |
Maximum Limit |
|---|---|
| Row Length | 1.5 MB |
| Number of columns | 4096 columns |
| Number of indexes | 128 |
| Total Number of Columns in a Single Index | 64 columns |
| Index Length | 16K |
| Total Number of Primary Key Columns | 64 columns |
| Primary Key Length | 16K |
| Number of Partitions |
|
| Maximum data volume supported by a single table | OceanBase Database theoretically supports an unlimited amount of data. According to best practices, a capacity of 50 million records is recommended for general scenarios. For wide tables with 100–200 columns, it is suggested to limit the data volume to around 10 million records. |
Single-column limits
Type |
Maximum Limit |
|---|---|
| Length of an index column | 16K |
String type limits
Type |
Maximum length |
|---|---|
CHAR |
256 characters |
VARCHAR |
262,144 characters |
BINARY |
256 bytes |
VARBINARY |
1,048,576 bytes |
TINYBLOB |
255 bytes |
BLOB |
65,535 bytes |
MEDIUMBLOB |
16,777,215 bytes |
LONGBLOB |
536,870,910 bytes |
TINYTEXT |
255 bytes |
TEXT |
65,535 bytes |
MEDIUMTEXT |
16,777,215 bytes |
LONGTEXT |
536,870,910 bytes |
Type |
Maximum length |
|---|---|
CHAR |
2,000 bytes |
NCHAR |
2,000 bytes |
VARCHAR |
32767 bytes |
VARCHAR2 |
32767 bytes |
NVARCHAR2 |
32767 bytes |
BLOB |
536870910 bytes |
CLOB |
536870910 bytes |
Transaction size limit
OceanBase Database theoretically supports unlimited transaction sizes, which are only limited by storage space.
Weak-consistency read latency limit
The weak-consistency read latency supported by OceanBase Database is controlled by the tenant-level parameter max_stale_time_for_weak_consistency. The default value is 5 seconds, meaning that with a minimum latency of 5 seconds, you can read the latest data.
Clog-related limitations
Type |
Limitations |
|---|---|
| Size of a single clog | 3.5 MB |
| Maximum Network Delay for Clog Synchronization | The maximum network latency for clog log synchronization is within 1 second, but the one-way network latency between the arbitration service and a node of the OceanBase cluster is less than or equal to 800 ms. |
Feature usage limitations
Shared storage
In the current version, for an OceanBase cluster in Shared-Storage (SS) mode, there are the following feature usage limitations:
- Multi-region deployment is not supported.
- The columnar storage engine is not supported.
- Physical standby databases are not supported.
- Reducing a tenant's data_disk_size is not supported.
- The number of partitions cannot exceed 100,000.
Physical standby database
The usage limitations for physical standby databases are listed in the following table.
Limitations |
Detailed Description |
|---|---|
| Maximum Number of Standby Tenants Supported by a Primary Tenant | There is no limit on the number of standby tenants that a primary tenant can support at the product level. However, since standby tenant synchronization consumes the primary tenant's CPU, memory, I/O resources, and network bandwidth, it is recommended to determine the number of standby tenants based on the source tenant's resource usage and actual business requirements. |
| Whether resource homogeneity is required for primary and standby tenants | Homogeneity is not required. It is recommended that the primary and standby tenants use the same resource specifications. |
| Parameter | The parameters of the primary tenant and standby tenant are independent of each other and are not physically synchronized. If you modify a parameter of the primary tenant, you must evaluate whether to modify the corresponding parameter of the standby tenant. |
| System Variables | System variables are physically synchronized between the primary and standby tenants. If you modify a system variable in the primary tenant, the system will synchronize the change to the corresponding system variable in the standby tenant. |
| User and Password | You can create users and change user passwords only on the primary tenant. The updated information is synchronized to the standby tenants. |
| Read and write operations | Standby tenants support read operations but do not support write operations. |
| Minor compaction and major compaction | Minor compactions of the primary tenant and standby tenants are relatively independent. Standby tenants synchronize major compaction information from the primary tenant and do not support independent major compaction. |
| Switchover Limitations | All replicas of the standby tenant's log stream must be online. |
| Failover Limitations | All replicas of the standby tenant's log stream must be online. |
| Optimizer statistics | Standby tenants do not collect statistics; their statistics are synchronized from the primary tenant. |
